Advanced search

Message boards : Number crunching : Unfettered use of GPUGrid

Author Message
Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 42998 - Posted: 14 Mar 2016 | 10:33:14 UTC

Gerard, has having unfettered use of GPUGrid helped your research?

Gerard
Send message
Joined: 26 Mar 14
Posts: 101
Credit: 0
RAC: 0
Level

Scientific publications
wat
Message 42999 - Posted: 14 Mar 2016 | 11:40:29 UTC - in response to Message 42998.
Last modified: 14 Mar 2016 | 12:18:23 UTC

Hi Betting Slip,

I am sorry I don't update you as much as I would want, but there's several reasons for this: mainly being in charge of updating GPUGRID while pursuing my PhD is too much work sometimes, secondarily some of the results cannot made public until published.

Your contribution to GPUGRID has helped a lot. The protocol developed in our lab (http://pubs.acs.org/doi/abs/10.1021/ct400919u) seems to work for a subset of molecules by which no binding pose was previously known. We are trying to improve inhibitors exploiting the knowledge generated. However, all these results are still theoretical and a second experimental phase, that we are currently undergoing, will be necessary before results can be made public.

Alternatively, for the past months I have been testing an old protocol (umbrella sampling) that seems to work like charm to get a quick idea about the binding affinity of ligands.

It is a very extense work, highly applicational and with real impact in biomedicine. It is not a mere proof of concept, we are really pushing the edges of knowdledge here, but slowly and patiently.

I hope this gave you a little insight (sorry to be succint) of the trascendence of your contribution. :)

Betting Slip
Send message
Joined: 5 Jan 09
Posts: 670
Credit: 2,498,095,550
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43000 - Posted: 14 Mar 2016 | 12:40:36 UTC - in response to Message 42999.

Thank you very much Gerard for taking time out to give all participants of this project your update. I'm particularly pleased your work is not just "proof of concept"

Thanks
Richard

[CSF] Thomas H.V. DUPONT
Send message
Joined: 20 Jul 14
Posts: 732
Credit: 115,270,366
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwat
Message 43001 - Posted: 14 Mar 2016 | 14:36:07 UTC - in response to Message 42999.

+1 :)
Thanks Gerard! Really appreciated!
____________
[CSF] Thomas H.V. Dupont
Founder of the team CRUNCHERS SANS FRONTIERES 2.0
www.crunchersansfrontieres

Jim1348
Send message
Joined: 28 Jul 12
Posts: 819
Credit: 1,591,285,971
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43003 - Posted: 14 Mar 2016 | 14:52:34 UTC

Great stuff. I keep one card on GPUGrid all the time, and will add more when the work increases. You can't do projects just to provide us work, it is important that you do something worthwhile. But it looks like the added capacity has helped you somewhat, which is very nice.

Vagelis Giannadakis
Send message
Joined: 5 May 13
Posts: 187
Credit: 349,254,454
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 43011 - Posted: 15 Mar 2016 | 10:28:17 UTC

Thank you Gerard for this information! I'm very happy to know we are contributing to some actual, real knowledge-expanding research here!
____________

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 0
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43013 - Posted: 15 Mar 2016 | 10:33:12 UTC - in response to Message 42999.

Thumbs up!

Skyler Baker
Send message
Joined: 19 Feb 16
Posts: 19
Credit: 140,656,383
RAC: 0
Level
Cys
Scientific publications
wat
Message 43105 - Posted: 29 Mar 2016 | 14:27:57 UTC - in response to Message 43013.

Thumbs up!



Hey you're that guy I'm chasing on project completion time :D you must have that stupid card LN2 cooled

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2343
Credit: 16,201,255,749
RAC: 0
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43108 - Posted: 29 Mar 2016 | 15:10:08 UTC - in response to Message 43105.
Last modified: 29 Mar 2016 | 15:10:31 UTC

Thumbs up!

Hey you're that guy I'm chasing on project completion time :D you must have that stupid card LN2 cooled

No, LN2 cooling for a 24/7 system is a complete economical nonsense. Perhaps some refrigerated water cooling could worth it.
All of my GPUs are standard air cooled ones. They are the fastest because of the lack of the infamous WDDM overhead in my systems, as I am using Windows XP x64 for crunching. That's it.

Skyler Baker
Send message
Joined: 19 Feb 16
Posts: 19
Credit: 140,656,383
RAC: 0
Level
Cys
Scientific publications
wat
Message 43109 - Posted: 29 Mar 2016 | 16:31:21 UTC - in response to Message 43108.

Thumbs up!

Hey you're that guy I'm chasing on project completion time :D you must have that stupid card LN2 cooled

No, LN2 cooling for a 24/7 system is a complete economical nonsense. Perhaps some refrigerated water cooling could worth it.
All of my GPUs are standard air cooled ones. They are the fastest because of the lack of the infamous WDDM overhead in my systems, as I am using Windows XP x64 for crunching. That's it.



I was joking about the LN2, I did figure they were probably heavily over clocked,interesting about the WDDM overhead. Wonder if there is any solution besides going to XP

Jim1348
Send message
Joined: 28 Jul 12
Posts: 819
Credit: 1,591,285,971
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43111 - Posted: 29 Mar 2016 | 21:15:05 UTC - in response to Message 43108.

No, LN2 cooling for a 24/7 system is a complete economical nonsense. Perhaps some refrigerated water cooling could worth it.
All of my GPUs are standard air cooled ones. They are the fastest because of the lack of the infamous WDDM overhead in my systems, as I am using Windows XP x64 for crunching. That's it.

I am sure you are right about the LN2. And I tried WinXP (32-bit) for a while, with good results. But the cards (probably GTX 660 as I recall) were a little less stable on WinXP than on Win7, and so I had to reduce the clocks a bit to eliminate the "unstable machine" notices. On the other hand, on Win7, I could overclock some, though I usually avoid it.

So my impression is that at the stability limit, you can get about the same results with Win7 as on WinXP, you just go about it differently. But I have not tried extensive tests, and have not used WinXP 64-bit at all.

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43119 - Posted: 30 Mar 2016 | 18:57:31 UTC - in response to Message 43111.

Wonder if there is any solution besides going to XP

It's called Linux (x64).
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 43120 - Posted: 30 Mar 2016 | 19:02:04 UTC - in response to Message 43111.
Last modified: 30 Mar 2016 | 19:02:16 UTC

So my impression is that at the stability limit, you can get about the same results with Win7 as on WinXP, you just go about it differently. But I have not tried extensive tests, and have not used WinXP 64-bit at all.

The bottom line is that WinXP and Linux allow you to get tasks done quicker. The fact that the card is used more/runs hotter is a result of not having the WDDM overhead. A better cooling strategy covers that.
With high end/enthusiast cards the WDDM overhead is probably higher.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Skyler Baker
Send message
Joined: 19 Feb 16
Posts: 19
Credit: 140,656,383
RAC: 0
Level
Cys
Scientific publications
wat
Message 43121 - Posted: 30 Mar 2016 | 22:17:33 UTC

Disabling hyperthreading also seemed to increase gpu usage

Post to thread

Message boards : Number crunching : Unfettered use of GPUGrid

//